To identify the key formulation factors controlling the initial drug and polymer dissolution rates from an amorphous solid dispersion (ASD). Ketoconazole (KTZ) ASDs using PVP, PVP-VA, HMPC, or HPMC-AS as polymeric matrix were prepared. For each drug-polymer system, two types of formulations with the same composition were prepared: 1. Spray dried dispersion (SDD) that is homogenous at molecular level, 2. Physical blend of SDD (80% drug loading) and pure polymer (SDD-PB) that is homogenous only at powder level. Flory-Huggins interaction parameters (χ) between KTZ and the four polymers were obtained by Flory-Huggins model fitting. Solution (13)C NMR and FT-IR were conducted to investigate the specific drug-polymer interaction in the solution and solid state, respectively. Intrinsic dissolution of both the drug and the polymer from ASDs were studied using a Higuchi style intrinsic dissolution apparatus. PXRD and confocal Raman microscopy were used to confirm the absence of drug crystallinity on the tablet surface before and after dissolution study. In solid state, KTZ is completely miscible with PVP, PVP-VA, or HPMC-AS, demonstrated by the negative χ values of -0.36, -0.46, -1.68, respectively; while is poorly miscible with HPMC shown by a positive χ value of 0.23. According to solution (13)C NMR and FT-IR studies, KTZ interacts with HPMC-AS strongly through H-bonding and dipole induced interaction; with PVPs and PVP-VA moderately through dipole-induced interactions; and with HPMC weakly without detectable attractive interaction. Furthermore, the "apparent" strength of drug-polymer interaction, measured by the extent of peak shift on NMR or FT-IR spectra, increases with the increasing number of interacting drug-polymer pairs. For ASDs with the presence of considerable drug-polymer interactions, such as KTZ/PVPs, KTZ/PVP-VA, or KTZ /HPMC-AS systems, drug released at the same rate as the polymer when intimate drug-polymer mixing was ensured (i.e., the SDD systems); while drug released much slower than the polymer when molecular level mixing or drug-polymer interaction was absent (SDD-PB systems). For ASDs without drug-polymer interaction (i.e., KTZ/HPMC systems), the mixing homogeneity had little impact on the release rate of either the drug or the polymer thus SDD and SDD-PB demonstrated the same drug or polymer release rate, while the drug released slowly and independently of polymer release. The initial drug release from an ASD was controlled by 1) the polymer release rate; 2) the strength of drug-polymer interaction, including the intrinsic interaction caused by the chemistry of the drug and the polymer (measured by the χ value), as well as that the apparent interaction caused by the drug-polymer ratio (measure by the extent of peak shift on spectroscopic analysis); and 3) the level of mixing homogeneity between the drug and polymer. In summary, the selection of polymer, drug-polymer ratio, and ASD processing conditions have profound impacts on the dissolution behavior of ASDs. Graphical Abstract Relationship between initial drug and polymer dissolution rates from amorphous solid dispersions with different mixing uniformity and drug-polymer interactions.